Re: [ACTION-208] Add voting support to Locationzation Quality Précis

Hi Phil,

Not sure what is gained by going to 1–100 decimal over 0–1 decimal (assuming that we don't have a fixed number of decimal places): 98.1234567891234567 is just as precise as 0.981234567891234567. In a sense it doesn't matter what the range is that we use as long as (a) we all agree on it and (b) we don't have a fixed float on the value. Just to be perverse we could use range values of 1.432 to 8.234 and we'd still have the same precision, although the math to convert it to something intelligible would add a little extra burden. But in general for computational processes it simplifies things to go from 0 to 1 since, assuming we equate 0 to 100 with percentage, you have to convert to the decimal fraction in some fashion for the math to work out anyway.

To take your example values, in the 0–100 space you would have this:

89.7, 89.9

But in the 0–1 you would have

0.897, 0.898

Both have three significant digits, so nether represents a loss in precision over the other.

But maybe I'm missing something here.

-Arle


On Aug 31, 2012, at 10:53 AM, Phil Ritchie <philr@vistatec.ie> wrote:

> My requirement for scores is that I need potentially hundreds of discrete values. For sake of interoperability I'm happy to map to a range but I don't want to end up "loosing precision" amongst my values as a consequence: i.e. 0-2000 of my scores map to 0-1 ITS and thus give values such as 0.897, 0.898, etc. 
> 
> My request would be to standardise on 0-100 decimal. 
> 
> Phil.
> 
> 
> 
> 
> 
> From:        Yves Savourel <ysavourel@enlaso.com> 
> To:        <public-multilingualweb-lt@w3.org>, 
> Date:        30/08/2012 21:59 
> Subject:        [ACTION-208] Add voting support to Locationzation Quality Précis 
> 
> 
> 
> Hi all,
> 
> I've added the attributes locQualityPrecisVote and locQualityPrecisVotePointer to the data category.
> http://www.w3.org/International/multilingualweb/lt/drafts/its20/its20.html#lqprecis
> 
> 
> - The name is not great maybe we should have locQualityPrecisVotingScore and locQualityPrecisRangeScore but this is getting just too long and probably confusing.
> 
> - I've adapted an example to show the voting, but a better one are certainly welcome.
> 
> - Currently we can use either vote or score not both. This allows to share the threshold. Hopefully this is fine. Otherwise we may have to also have a threshold for the voting.
> 
> - We've talked about harmonizing the scores/ranges in general. Currently locQualityPrecisScore is an integer between 0 and 100 (inclusive). Should we move to a decimal between 0 and 1?
> 
> Cheers,
> -yves
> 
> 
> 
> 
> 
> ************************************************************
> This email and any files transmitted with it are confidential and
> intended solely for the use of the individual or entity to whom they
> are addressed. If you have received this email in error please notify
> the sender immediately by e-mail.
> 
> www.vistatec.com
> ************************************************************
>